Deliverable 4.1 of the UNTWIST project: UNTWIST Gender sub-scheme and coding-book instructions

This document consists in a report on the coding procedure for party manifestos, the first main deliverable of Work Package 4 (WP4) of the Untwist project. WP4 is concerned with mapping feminist permeation in party manifestos, and estimating how parties in the six UNTWIST country-contexts - Denmark, Germany, Hungary, Spain, Switzerland, and the United Kingdom - address gender-based needs (GBNs). Analysing whether gender-based needs are covered in party manifestos is a fundamental step for assessing areas of weak and strong, sufficient, or insufficient representation of GBNs by parties. On the one hand, this research effort will help to explain and expose which GBNs have been neglected by mainstream politics, thus allowing us to identify areas where representation can and should be increased. On the other hand, mapping GBNs in party manifestos will help to expose areas where these needs have been addressed in a twisted or untwisted fashion by specific political actors, in particular radical-right parties.

The purpose of this document is to first present the background of this research endeavour, and the reasons justifying to the development of an original coding procedure (Section 2). The analysis of party manifestos in political science involves studying party preferences and behaviour by quantitatively analysing electoral programs. This method, established through projects like Manifesto Research on Political Representation (MARPOR) and Euromanifesto (EM), involves expert coders breaking down manifestos into predefined categories to understand issues parties prioritise and their positions about these topics. While this approach presents limitations, its strengths outweigh its weaknesses. Existing procedures explicitly dedicated to GBNs fail to thoroughly address these topics in party politics. New coding schemes, like those in the EM and Regional Manifesto (RMP) projects, attempt to capture gender-related issues but face limitations due to broad categories and forcing a choice between gender topics and others. Building on this background, WP4 developed a new coding procedure for a more comprehensive analysis of GBNs in party manifestos.

The document continues by describing the data collection, the development of the new coding procedure, their link with the Typology developed by WP1 (D1.1), and how these pave the way for developing machine-learning methods for the computational annotation of party manifestos (Section 3). In pursuit of a comprehensive Manifesto Gender Analysis (MGA), WP4 integrated pre-existing party manifestos from MARPOR and EM projects across six national contexts from 2004 to 2021, yielding 453 diverse documents in various formats. The MGA procedure unfolds alongside two key elements: the annotation of manifestos through quasi-sentence splitting and coding based on a novel scheme, and an accompanying expert survey capturing nuances beyond isolated sentences. Iteratively revised within WP4 and the UNTWIST consortium, the MGA's coding categories drew from diverse sources, notably aligning with WP1's gender typology. This fine-grained yet scalable procedure, while complex, fosters both broad and detailed analyses of how parties address GBNs, primed for potential computational advancements in language processing, offering cost-effective expansion possibilities. Moreover, its hierarchical annotation structure paves the way for robustness checks and refined measurement methods, potentially overcoming limitations of single-coder reliance in manual procedures.

Finally, the document provides a summary of the MGA coding procedure (Section 4).